refactor: move all state to struct, add footnotes handling

This commit is contained in:
DarkCat09 2024-11-21 22:43:08 +04:00
parent 959581d0b3
commit 852aabd1b2
Signed by: DarkCat09
GPG key ID: BD3CE9B65916CD82

View file

@ -12,13 +12,17 @@ pub fn compile_markdown(
) -> std::io::Result<()> { ) -> std::io::Result<()> {
let src_text = std::fs::read_to_string(src)?; let src_text = std::fs::read_to_string(src)?;
let mut html = create_file(html)?; let mut p = ParserVars {
let mut gmi = create_file(gmi)?; html: Box::new(create_file(html)?),
gmi: Box::new(create_file(gmi)?),
let mut state = State::Start; state: State::Start,
let mut ol = false; // ordered list ol: false,
let mut counter: u64 = 0; counter: 0,
let mut links: Vec<GmiLink<'_>> = vec![];
links: vec![],
notes: vec![],
};
for event in Parser::new_ext(&src_text, Options::all()) { for event in Parser::new_ext(&src_text, Options::all()) {
use pulldown_cmark::Event::*; use pulldown_cmark::Event::*;
@ -31,143 +35,145 @@ pub fn compile_markdown(
match tag { match tag {
Paragraph => { Paragraph => {
html.write_all(b"<p>")?; p.html.write_all(b"<p>")?;
write_paragraph_start(&mut gmi, state)?; write_paragraph_start(&mut p)?;
} }
Heading { level, id, .. } => { Heading { level, id, .. } => {
if let Some(id) = id { if let Some(id) = id {
html.write_fmt(format_args!("<{} id=\"{}\">", level, id))?; p.html
.write_fmt(format_args!("<{} id=\"{}\">", level, id))?;
} else { } else {
html.write_fmt(format_args!("<{}>", level))?; p.html.write_fmt(format_args!("<{}>", level))?;
} }
write_paragraph_start(&mut gmi, state)?; write_paragraph_start(&mut p)?;
let hashes = match level { let hashes = match level {
HeadingLevel::H1 => "# ", HeadingLevel::H1 => "# ",
HeadingLevel::H2 => "## ", HeadingLevel::H2 => "## ",
_ => "### ", _ => "### ",
}; };
gmi.write_all(hashes.as_bytes())?; p.gmi.write_all(hashes.as_bytes())?;
state = State::Paragraph; p.state = State::Paragraph;
} }
BlockQuote(_kind) => { BlockQuote(_kind) => {
html.write_all(b"<blockquote>")?; p.html.write_all(b"<blockquote>")?;
write_paragraph_start(&mut gmi, state)?; write_paragraph_start(&mut p)?;
gmi.write_all(b"> ")?; p.gmi.write_all(b"> ")?;
state = State::Quote; p.state = State::Quote;
} }
CodeBlock(CodeBlockKind::Fenced(lang)) => { CodeBlock(CodeBlockKind::Fenced(lang)) => {
// TODO: highlighting with syntect // TODO: highlighting with syntect
html.write_all(b"<pre><code>")?; p.html.write_all(b"<pre><code>")?;
write_paragraph_start(&mut gmi, state)?; write_paragraph_start(&mut p)?;
gmi.write_fmt(format_args!("```{}\r\n", lang))?; p.gmi.write_fmt(format_args!("```{}\r\n", lang))?;
} }
CodeBlock(CodeBlockKind::Indented) => { CodeBlock(CodeBlockKind::Indented) => {
html.write_all(b"<pre><code>")?; p.html.write_all(b"<pre><code>")?;
write_paragraph_start(&mut gmi, state)?; write_paragraph_start(&mut p)?;
gmi.write_all(b"```")?; p.gmi.write_all(b"```")?;
} }
List(None) => { List(None) => {
html.write_all(b"<ul>")?; p.html.write_all(b"<ul>")?;
write_paragraph_start(&mut gmi, state)?; write_paragraph_start(&mut p)?;
state = State::Start; p.state = State::Start;
} }
List(Some(counter_start)) => { List(Some(counter_start)) => {
counter = counter_start; p.counter = counter_start;
ol = true; p.ol = true;
if counter_start == 1 { if counter_start == 1 {
html.write_all(b"<ol>")?; p.html.write_all(b"<ol>")?;
} else { } else {
html.write_fmt(format_args!("<ol start=\"{}\">", counter_start))?; p.html
.write_fmt(format_args!("<ol start=\"{}\">", counter_start))?;
} }
write_paragraph_start(&mut gmi, state)?; write_paragraph_start(&mut p)?;
state = State::Start; p.state = State::Start;
} }
Item => { Item => {
html.write_all(b"<li>")?; p.html.write_all(b"<li>")?;
if state != State::Start { if p.state != State::Start {
gmi.write_all(b"\r\n")?; p.gmi.write_all(b"\r\n")?;
state = State::Paragraph; p.state = State::Paragraph;
} }
if ol { if p.ol {
gmi.write_fmt(format_args!("{}. ", counter))?; p.gmi.write_fmt(format_args!("{}. ", p.counter))?;
counter += 1; p.counter += 1;
} else { } else {
gmi.write_all(b"* ")?; p.gmi.write_all(b"* ")?;
} }
} }
Strong => { Strong => {
html.write_all(b"<b>")?; p.html.write_all(b"<b>")?;
write_inline(&mut gmi, state, "**", &mut links)?; write_inline(&mut p, "**")?;
} }
Emphasis => { Emphasis => {
html.write_all(b"<i>")?; p.html.write_all(b"<i>")?;
write_inline(&mut gmi, state, "*", &mut links)?; write_inline(&mut p, "*")?;
} }
Strikethrough => { Strikethrough => {
html.write_all(b"<s>")?; p.html.write_all(b"<s>")?;
write_inline(&mut gmi, state, "~", &mut links)?; write_inline(&mut p, "~")?;
} }
Link { dest_url, .. } => { Link { dest_url, .. } => {
html.write_fmt(format_args!("<a href=\"{}\">", dest_url))?; p.html
links.push(GmiLink { .write_fmt(format_args!("<a href=\"{}\">", dest_url))?;
title: String::new(), p.links.push(GmiLink::from_url(dest_url));
url: dest_url, p.state = State::Link;
});
state = State::Link;
} }
Image { dest_url, .. } => { Image { dest_url, .. } => {
links.push(GmiLink { p.links.push(GmiLink::from_url(dest_url));
title: String::new(), p.state = State::Link;
url: dest_url,
});
state = State::Link;
} }
Table(_align) => { Table(_align) => {
html.write_all(b"<table>")?; p.html.write_all(b"<table>")?;
write_paragraph_start(&mut gmi, state)?; write_paragraph_start(&mut p)?;
// gmi ?? // gmi ??
} }
TableHead => { TableHead => {
html.write_all(b"<thead><tr>")?; p.html.write_all(b"<thead><tr>")?;
state = State::TableHead; p.state = State::TableHead;
} }
TableRow => { TableRow => {
html.write_all(b"<tr>")?; p.html.write_all(b"<tr>")?;
} }
TableCell => { TableCell => {
if state == State::TableHead { if p.state == State::TableHead {
html.write_all(b"<th>")?; p.html.write_all(b"<th>")?;
} else { } else {
html.write_all(b"<td>")?; p.html.write_all(b"<td>")?;
} }
} }
FootnoteDefinition(id) => {
p.notes.push(Footnote::from_id(id));
p.state = State::Footnote;
}
MetadataBlock(_) => { MetadataBlock(_) => {
state = State::Metadata(state != State::Start); p.state = State::Metadata(p.state != State::Start);
} }
other => { other => {
@ -181,67 +187,67 @@ pub fn compile_markdown(
match tag { match tag {
Paragraph => { Paragraph => {
html.write_all(b"</p>")?; p.html.write_all(b"</p>")?;
if !links.is_empty() { if !p.links.is_empty() {
gmi.write_all(b"\r\n")?; p.gmi.write_all(b"\r\n")?;
for (i, link) in links.iter().enumerate() { for (i, link) in p.links.iter().enumerate() {
gmi.write_fmt(format_args!( p.gmi.write_fmt(format_args!(
"\r\n=> {} [{}]: {}", // => https://... [1]: example "\r\n=> {} [{}]: {}", // => https://... [1]: example
link.url, i, link.title, link.url, i, link.title,
))?; ))?;
} }
links.clear(); p.links.clear();
} }
} }
Heading(level) => html.write_fmt(format_args!("<{}>", level))?, Heading(level) => p.html.write_fmt(format_args!("<{}>", level))?,
BlockQuote(_) => html.write_all(b"</blockquote>")?, BlockQuote(_) => p.html.write_all(b"</blockquote>")?,
CodeBlock => { CodeBlock => {
html.write_all(b"</code></pre>")?; p.html.write_all(b"</code></pre>")?;
gmi.write_all(b"```")?; p.gmi.write_all(b"```")?;
} }
List(ordered) => { List(ordered) => {
if ordered { if ordered {
html.write_all(b"</ol>")?; p.html.write_all(b"</ol>")?;
ol = false; p.ol = false;
} else { } else {
html.write_all(b"</ul>")?; p.html.write_all(b"</ul>")?;
} }
} }
Item => html.write_all(b"</li>")?, Item => p.html.write_all(b"</li>")?,
Strong => { Strong => {
html.write_all(b"</b>")?; p.html.write_all(b"</b>")?;
write_inline(&mut gmi, state, "**", &mut links)?; write_inline(&mut p, "**")?;
} }
Emphasis => { Emphasis => {
html.write_all(b"</i>")?; p.html.write_all(b"</i>")?;
write_inline(&mut gmi, state, "*", &mut links)?; write_inline(&mut p, "*")?;
} }
Strikethrough => { Strikethrough => {
html.write_all(b"</s>")?; p.html.write_all(b"</s>")?;
write_inline(&mut gmi, state, "~", &mut links)?; write_inline(&mut p, "~")?;
} }
Link | Image => { Link | Image => {
gmi.write_fmt(format_args!( p.gmi.write_fmt(format_args!(
"{}[{}]", // example[1] ...\r\n => https://... [1]: example "{}[{}]", // example[1] ...\r\n => https://... [1]: example
links.last().unwrap().title, p.links.last().unwrap().title,
links.len(), p.links.len(),
))?; ))?;
} }
TableHead => html.write_all(b"</tr></thead><tbody>")?, TableHead => p.html.write_all(b"</tr></thead><tbody>")?,
TableRow => html.write_all(b"</tr>")?, TableRow => p.html.write_all(b"</tr>")?,
TableCell => { TableCell => {
if state == State::TableHead { if p.state == State::TableHead {
html.write_all(b"</th>")? p.html.write_all(b"</th>")?
} else { } else {
html.write_all(b"</td>")? p.html.write_all(b"</td>")?
} }
} }
Table => html.write_all(b"</tbody></table>")?, Table => p.html.write_all(b"</tbody></table>")?,
_ => {} _ => {}
} }
state = if state == State::Metadata(false) { p.state = if p.state == State::Metadata(false) {
State::Start State::Start
} else { } else {
State::Paragraph State::Paragraph
@ -249,7 +255,7 @@ pub fn compile_markdown(
} }
Text(text) => { Text(text) => {
match state { match p.state {
State::Metadata(_) => { State::Metadata(_) => {
// TODO: parse yaml // TODO: parse yaml
continue; continue;
@ -257,48 +263,61 @@ pub fn compile_markdown(
_ => {} _ => {}
} }
html.write_all(text.as_bytes())?; p.html.write_all(text.as_bytes())?;
write_inline(&mut gmi, state, &text, &mut links)?; write_inline(&mut p, &text)?;
} }
Code(code) => { Code(code) => {
html.write_all(b"<code>")?; p.html.write_all(b"<code>")?;
html.write_all(code.as_bytes())?; p.html.write_all(code.as_bytes())?;
html.write_all(b"</code>")?; p.html.write_all(b"</code>")?;
write_inline(&mut gmi, state, "`", &mut links)?; write_inline(&mut p, "`")?;
write_inline(&mut gmi, state, &code, &mut links)?; write_inline(&mut p, &code)?;
write_inline(&mut gmi, state, "`", &mut links)?; write_inline(&mut p, "`")?;
} }
SoftBreak => { SoftBreak => {
html.write_all(b" ")?; p.html.write_all(b" ")?;
write_inline(&mut gmi, state, " ", &mut links)?; write_inline(&mut p, " ")?;
} }
HardBreak => { HardBreak => {
html.write_all(b"<br>")?; p.html.write_all(b"<br>")?;
gmi.write_all(b"\r\n")?; write_inline(&mut p, "\r\n")?;
if state == State::Quote { if p.state == State::Quote {
gmi.write_all(b"> ")?; p.gmi.write_all(b"> ")?;
} }
} }
Rule => { Rule => {
html.write_all(b"<hr>")?; p.html.write_all(b"<hr>")?;
write_paragraph_start(&mut gmi, state)?; write_paragraph_start(&mut p)?;
gmi.write_all(b"---")?; p.gmi.write_all(b"---")?;
}
FootnoteReference(id) => {
p.html.write_fmt(format_args!(
"<sup><a href=\"#ref-{}\">[{}]</a></sup>",
id, id,
))?;
// [^1]
p.gmi.write_all(b"[^")?;
p.gmi.write_all(id.as_bytes())?;
p.gmi.write_all(b"]")?;
} }
TaskListMarker(done) => { TaskListMarker(done) => {
if done { if done {
html.write_all(b"<input type=checkbox checked disabled>")?; p.html
gmi.write_all(b"[x] ")?; .write_all(b"<input type=checkbox checked disabled>")?;
p.gmi.write_all(b"[x] ")?;
} else { } else {
html.write_all(b"<input type=checkbox disabled>")?; p.html.write_all(b"<input type=checkbox disabled>")?;
gmi.write_all(b"[ ] ")?; p.gmi.write_all(b"[ ] ")?;
} }
} }
@ -308,8 +327,10 @@ pub fn compile_markdown(
} }
} }
html.flush()?; // TODO: footnotes rendering
gmi.flush()?;
p.html.flush()?;
p.gmi.flush()?;
Ok(()) Ok(())
} }
@ -324,33 +345,50 @@ fn create_file(path: impl AsRef<Path>) -> std::io::Result<BufWriter<std::fs::Fil
} }
#[inline(always)] #[inline(always)]
fn write_paragraph_start(gmi: &mut impl std::io::Write, state: State) -> std::io::Result<()> { fn write_paragraph_start(p: &mut ParserVars) -> std::io::Result<()> {
if state != State::Start && state != State::Quote { match p.state {
gmi.write_all(b"\r\n\r\n")?; State::Start | State::Quote => {}
_ => {
p.gmi.write_all(b"\r\n\r\n")?;
}
} }
Ok(()) Ok(())
} }
#[inline(always)] #[inline(always)]
fn write_inline( fn write_inline(p: &mut ParserVars, text: &str) -> std::io::Result<()> {
gmi: &mut impl std::io::Write, match p.state {
state: State, State::Link => {
text: &str, p.links.last_mut().unwrap().title.push_str(text);
links: &mut Vec<GmiLink>, }
) -> std::io::Result<()> { State::Footnote => {
if state == State::Link { p.notes.last_mut().unwrap().content.push_str(text);
links.last_mut().unwrap().title.push_str(text); }
} else { _ => {
gmi.write_all(text.as_bytes())?; p.gmi.write_all(text.as_bytes())?;
}
} }
Ok(()) Ok(())
} }
struct ParserVars<'v> {
html: Box<dyn std::io::Write>,
gmi: Box<dyn std::io::Write>,
state: State,
ol: bool,
counter: u64,
links: Vec<GmiLink<'v>>,
notes: Vec<Footnote<'v>>,
}
#[derive(Debug, Clone, Copy, PartialEq, Eq)] #[derive(Debug, Clone, Copy, PartialEq, Eq)]
enum State { enum State {
Start, Start,
Paragraph, Paragraph,
Link, Link,
Footnote,
Quote, Quote,
TableHead, TableHead,
Metadata(bool), Metadata(bool),
@ -360,3 +398,26 @@ struct GmiLink<'link> {
title: String, title: String,
url: CowStr<'link>, url: CowStr<'link>,
} }
impl<'link> GmiLink<'link> {
fn from_url(value: CowStr<'link>) -> Self {
Self {
title: String::new(),
url: value,
}
}
}
struct Footnote<'note> {
id: CowStr<'note>,
content: String,
}
impl<'note> Footnote<'note> {
fn from_id(value: CowStr<'note>) -> Self {
Self {
id: value,
content: String::new(),
}
}
}